【摘要】 论文研究了汉语小词汇表语音识别算法的基本原理,提出了具有鲁棒性的两级端点检测语音识别技术,在语音信号采集时,根据过零率、短时能量对数据进行提取并压缩,采用了多模板匹配算法识别。硬件采用51内核单片机,用较少的存储空间和计算空间实现语音数据处理,不需要额外的器件。实验用20个字的汉语小词汇量系统进行了测试,识别成功率大于90%,显示该算法比通常采用的算法性能更好。更多还原

【Abstract】 This paper introduces the basic principles of Chinese small vocabularies speech recognition algorithm.Two end-point processing techniques for robust speech recognition are presented and templates are made after performing data extractions and compressions.The input pattern obtained during run time is classified according to the zero crossing rate,the short-time energy.Many level end-point detections are adopted in the system.The recognition system,based on a 51 single chip to operate in real-time as well as with the lowest possible computational and memory costs,does not require external devices.A system evaluation using a vocabulary of 20 words shows a success rate of more than 90%,demonstrating that the algorithm having much better performance than that of the commonly adopted algorithm.更多还原